The paper proposes a theoretical framework on migrant workers' choice of jobs and places of employment. It argues that because of urban-rural segmentation in the labor market and extremely high mobility costs, migrant workers' choice-making is driven mainly by household utility maximization, rather than by salary maximization. As a result, there may be biases in estimating the educational return for migrant workers using the traditional Mincer wage equation, which relies on the efficient labor market assump...